Merge branch 'master' into 10081-cwl-run-same-job

[arvados.git] / sdk / cwl / arvados_cwl / arvworkflow.py

diff --git a/sdk/cwl/arvados_cwl/arvworkflow.py b/sdk/cwl/arvados_cwl/arvworkflow.py
index 8e45890a8d6226d9387524e1dc330effe900e163..ab3940998bd669c3f5bc389866b19effb9a708d3 100644 (file)
--- a/sdk/cwl/arvados_cwl/arvworkflow.py
+++ b/sdk/cwl/arvados_cwl/arvworkflow.py
@@ -1,16 +1,22 @@
 import os
 import json
 import copy
+import logging
 
 from cwltool.pack import pack
 from cwltool.load_tool import fetch_document
 from cwltool.process import shortname
+from cwltool.workflow import Workflow
+from cwltool.pathmapper import adjustFileObjs, adjustDirObjs
 
 import ruamel.yaml as yaml
 
 from .runner import upload_docker, upload_dependencies
+from .arvtool import ArvadosCommandTool
 
-def make_workflow(arvRunner, tool, job_order, project_uuid, update_uuid):
+logger = logging.getLogger('arvados.cwl-runner')
+
+def upload_workflow(arvRunner, tool, job_order, project_uuid, update_uuid):
     upload_docker(arvRunner, tool)
 
     document_loader, workflowobj, uri = (tool.doc_loader, tool.doc_loader.fetch(tool.tool["id"]), tool.tool["id"])
@@ -39,3 +45,76 @@ def make_workflow(arvRunner, tool, job_order, project_uuid, update_uuid):
         return arvRunner.api.workflows().update(uuid=update_uuid, body=body).execute(num_retries=arvRunner.num_retries)["uuid"]
     else:
         return arvRunner.api.workflows().create(body=body).execute(num_retries=arvRunner.num_retries)["uuid"]
+
+class ArvadosWorkflow(Workflow):
+    """Wrap cwltool Workflow to override selected methods."""
+
+    def __init__(self, arvrunner, toolpath_object, **kwargs):
+        super(ArvadosWorkflow, self).__init__(toolpath_object, **kwargs)
+        self.arvrunner = arvrunner
+        self.work_api = kwargs["work_api"]
+
+    def job(self, joborder, output_callback, **kwargs):
+        kwargs["work_api"] = self.work_api
+        req, _ = self.get_requirement("http://arvados.org/cwl#RunInSingleContainer")
+        if req:
+            document_loader, workflowobj, uri = (self.doc_loader, self.doc_loader.fetch(self.tool["id"]), self.tool["id"])
+
+            workflowobj["requirements"] = self.requirements + workflowobj.get("requirements", [])
+            workflowobj["hints"] = self.hints + workflowobj.get("hints", [])
+            packed = pack(document_loader, workflowobj, uri, self.metadata)
+
+            upload_dependencies(self.arvrunner,
+                                kwargs.get("name", ""),
+                                document_loader,
+                                packed,
+                                uri,
+                                False)
+
+            upload_dependencies(self.arvrunner,
+                                os.path.basename(joborder.get("id", "#")),
+                                document_loader,
+                                joborder,
+                                joborder.get("id", "#"),
+                                False)
+
+            joborder_keepmount = copy.deepcopy(joborder)
+
+            def keepmount(obj):
+                if obj["location"].startswith("keep:"):
+                    obj["location"] = "/keep/" + obj["location"][5:]
+                else:
+                    raise Exception("Uh oh %s" % obj["location"])
+                if "listing" in obj:
+                    del obj["listing"]
+            adjustFileObjs(joborder_keepmount, keepmount)
+            adjustDirObjs(joborder_keepmount, keepmount)
+            adjustFileObjs(packed, keepmount)
+            adjustDirObjs(packed, keepmount)
+
+            wf_runner = {
+                "class": "CommandLineTool",
+                "baseCommand": "cwltool",
+                "inputs": self.tool["inputs"],
+                "outputs": self.tool["outputs"],
+                "stdout": "cwl.output.json",
+                "requirements": workflowobj["requirements"]+[
+                    {"class": "InlineJavascriptRequirement"},
+                    {
+                    "class": "InitialWorkDirRequirement",
+                    "listing": [{
+                            "entryname": "workflow.cwl",
+                            "entry": yaml.safe_dump(packed).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
+                        }, {
+                            "entryname": "cwl.input.json",
+                            "entry": yaml.safe_dump(joborder_keepmount).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
+                        }]
+                }],
+                "hints": workflowobj["hints"],
+                "arguments": ["--no-container", "--move-outputs", "workflow.cwl#main", "cwl.input.json"]
+            }
+            kwargs["loader"] = self.doc_loader
+            kwargs["avsc_names"] = self.doc_schema
+            return ArvadosCommandTool(self.arvrunner, wf_runner, **kwargs).job(joborder, output_callback, **kwargs)
+        else:
+            return super(ArvadosWorkflow, self).job(joborder, output_callback, **kwargs)